From 3bdbeef5d611a11ff425d3c3f415bd7bec855de7 Mon Sep 17 00:00:00 2001
From: Nathan Gray <nathangray.bsc@gmail.com>
Date: Wed, 7 Sep 2011 15:51:33 +0000
Subject: [PATCH] Taming tidy - fixes extra line breaks, don't detect encoding.

---
 etemplate/inc/class.bo_merge.inc.php      | 22 ++++++++++++++++++---
 etemplate/templates/default/msoffice.xslt | 24 ++++++++++++++++++++---
 etemplate/templates/default/wordml.xslt   | 18 +++++++++++++++--
 3 files changed, 56 insertions(+), 8 deletions(-)

diff --git a/etemplate/inc/class.bo_merge.inc.php b/etemplate/inc/class.bo_merge.inc.php
index 2edab94c90..13f4e39ae6 100644
--- a/etemplate/inc/class.bo_merge.inc.php
+++ b/etemplate/inc/class.bo_merge.inc.php
@@ -64,6 +64,12 @@ abstract class bo_merge
 		'clean'	=> true,
 		'output-xhtml'	=> true,
 		'show-body-only'	=> true,
+		'output-encoding'	=> 'utf-8',
+		'input-encoding'	=> 'utf-8',
+		'quote-ampersand'	=> false,	// Prevent double encoding
+		'quote-nbsp'		=> true,	// XSLT can handle spaces easier
+		'preserve-entities'	=> true,
+		'wrap'			=> 0,		// Wrapping can break output
 	);
 
 	/**
@@ -734,9 +740,19 @@ abstract class bo_merge
 				if (is_string($value) && (strpos($value,'<') !== false))
 				{
 					// Clean HTML, if it's being kept
-					if($replace_tags && extension_loaded('tidy'))
-					{
-						$value = tidy_repair_string($value, self::$tidy_config, 'utf8');
+					if($replace_tags && extension_loaded('tidy')) {
+						$tidy = new tidy();
+						$cleaned = $tidy->repairString($value, self::$tidy_config);
+						// Found errors. Strip it all so there's some output
+						if($tidy->getStatus() == 2)
+						{
+							error_log($tidy->errorBuffer);
+							$value = strip_tags($value);
+						}
+						else
+						{
+							$value = $cleaned;
+						}
 					}
 					// replace </p> and <br /> with CRLF (remove <p> and CRLF)
 					$value = str_replace(array("\r","\n",'<p>','</p>','<br />'),array('','','',"\r\n","\r\n"),$value);
diff --git a/etemplate/templates/default/msoffice.xslt b/etemplate/templates/default/msoffice.xslt
index ebee7ad9dc..d39af698c1 100644
--- a/etemplate/templates/default/msoffice.xslt
+++ b/etemplate/templates/default/msoffice.xslt
@@ -86,7 +86,11 @@ Breakers
 	</xsl:template>
 -->
 
-	<xsl:template name="apply-styles" match="w:r[descendant::strong|descendant::em|descendant::u|descendant::span]">
+	<xsl:template match="w:r[descendant::strong|descendant::em|descendant::u|descendant::span]">
+		<xsl:call-template name="apply-styles"/>
+	</xsl:template>
+
+	<xsl:template name="apply-styles">
 		<xsl:for-each select="node()|@*[not(w:rPr)]">
 			<xsl:choose>
 			<xsl:when test="descendant::strong|descendant::em|descendant::u|descendant::span" >
@@ -210,9 +214,23 @@ Breakers
 					<w:t>&#xB7;</w:t><w:tab/>
 				</xsl:otherwise>
 			</xsl:choose>
-			</w:r><w:r>
-				<w:t><xsl:value-of select="normalize-space(text())" /></w:t>
 			</w:r>
+			<xsl:choose>
+				<xsl:when test="count(child::*)=0">
+					<xsl:variable name="text">
+						<xsl:value-of select="substring-after(text(),' ')"/>
+					</xsl:variable>
+					<w:r><w:t>
+<xsl:value-of select="normalize-space($text)"/>
+					</w:t></w:r>
+				</xsl:when>
+				<xsl:otherwise>
+					<!-- Strip out styles, they would need to be processed-->
+					<w:r><w:t>
+						<xsl:value-of select = "normalize-space(child::*)"/>
+					</w:t></w:r>
+				</xsl:otherwise>
+			</xsl:choose>
 			</w:p>
 		</xsl:for-each>
 	</xsl:template>
diff --git a/etemplate/templates/default/wordml.xslt b/etemplate/templates/default/wordml.xslt
index 15025c4255..d4be265251 100644
--- a/etemplate/templates/default/wordml.xslt
+++ b/etemplate/templates/default/wordml.xslt
@@ -208,9 +208,23 @@ Breakers
 				<w:t>&#xB7;</w:t><w:tab/>
 				</xsl:otherwise>
 			</xsl:choose>
-			</w:r><w:r>
-				<w:t><xsl:value-of select="normalize-space(text())" /></w:t>
 			</w:r>
+			<xsl:choose>
+				<xsl:when test="count(child::*)=0">
+					<xsl:variable name="text">
+						<xsl:value-of select="substring-after(text(),' ')"/>
+					</xsl:variable>
+					<w:r><w:t>
+<xsl:value-of select="normalize-space($text)"/>
+					</w:t></w:r>
+				</xsl:when>
+				<xsl:otherwise>
+					<!-- Strip out styles, they would need to be processed-->
+					<w:r><w:t>
+						<xsl:value-of select = "normalize-space(child::*)"/>
+					</w:t></w:r>
+				</xsl:otherwise>
+			</xsl:choose>
 			</w:p>
 		</xsl:for-each>
 	</xsl:template>